Optimizing 3D object detection for embedded systems in automated vehicles using sensor data fusion and CUDA computing

Topi Miekkala; Matti Kutila; Mathias Schneider; Alfred Höß

doi:10.1109/ICCP56966.2022.10053969

Optimizing 3D object detection for embedded systems in automated vehicles using sensor data fusion and CUDA computing

Topi Miekkala, Matti Kutila, Mathias Schneider, Alfred Höß

Ostbayerische Technische Hochschule Amberg-Weiden

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

1 Citation (Scopus)

239 Downloads (Pure)

Abstract

This article explores the utilization of the processing power of GPUs using CUDA computation for real-time aggregation of multi-sensor data and detection of 3D objects using parallel clustering algorithms. The purpose is to implement an algorithm that fuses raw lidar point cloud data and 2D camera image object detections to produce 3D object clusters in a lidar point cloud. Most of the computation has been implemented using CUDA parallelism to investigate the capability of GPU devices in this task, which is a common challenge in automated driving. The results indicate that processing times can be optimized within the algorithm, which is crucial when considering the large amounts of data provided by lidar and camera-based systems. The algorithm can perform inference on the Jetson Xavier AGX at rates of ~20 to ~220 ms depending on the number of objects and their corresponding point amounts in the KITTI dataset.

Original language	English
Title of host publication	Proceedings - 2022 IEEE 18th International Conference on Intelligent Computer Communication and Processing Conference, ICCP 2022
Editors	Sergiu Nedevschi, Rodica Potolea, Radu Razvan Slavescu
Publisher	IEEE Institute of Electrical and Electronic Engineers
Pages	113-118
Number of pages	6
ISBN (Electronic)	978-1-6654-6437-6
DOIs	https://doi.org/10.1109/ICCP56966.2022.10053969
Publication status	Published - 2022
MoE publication type	A4 Article in a conference publication
Event	2022 IEEE 18th International Conference on Intelligent Computer Communication and Processing, ICCP 2022 - Cluj-Napoca, Romania Duration: 22 Sept 2022 → 24 Sept 2022

Conference

Conference	2022 IEEE 18th International Conference on Intelligent Computer Communication and Processing, ICCP 2022
Country/Territory	Romania
City	Cluj-Napoca
Period	22/09/22 → 24/09/22

Keywords

3D object detection
automated driving
CUDA
sensor data fusion

Access to Document

10.1109/ICCP56966.2022.10053969

ICCP22_PAPER_7Accepted author manuscript, 1.29 MB

Cite this

Miekkala, T., Kutila, M., Schneider, M., & Höß, A. (2022). Optimizing 3D object detection for embedded systems in automated vehicles using sensor data fusion and CUDA computing. In S. Nedevschi, R. Potolea, & R. R. Slavescu (Eds.), Proceedings - 2022 IEEE 18th International Conference on Intelligent Computer Communication and Processing Conference, ICCP 2022 (pp. 113-118). IEEE Institute of Electrical and Electronic Engineers. https://doi.org/10.1109/ICCP56966.2022.10053969

Miekkala, Topi ; Kutila, Matti ; Schneider, Mathias et al. / Optimizing 3D object detection for embedded systems in automated vehicles using sensor data fusion and CUDA computing. Proceedings - 2022 IEEE 18th International Conference on Intelligent Computer Communication and Processing Conference, ICCP 2022. editor / Sergiu Nedevschi ; Rodica Potolea ; Radu Razvan Slavescu. IEEE Institute of Electrical and Electronic Engineers, 2022. pp. 113-118

@inproceedings{77f96e22b278471aba86340b3a8d7799,

title = "Optimizing 3D object detection for embedded systems in automated vehicles using sensor data fusion and CUDA computing",

abstract = "This article explores the utilization of the processing power of GPUs using CUDA computation for real-time aggregation of multi-sensor data and detection of 3D objects using parallel clustering algorithms. The purpose is to implement an algorithm that fuses raw lidar point cloud data and 2D camera image object detections to produce 3D object clusters in a lidar point cloud. Most of the computation has been implemented using CUDA parallelism to investigate the capability of GPU devices in this task, which is a common challenge in automated driving. The results indicate that processing times can be optimized within the algorithm, which is crucial when considering the large amounts of data provided by lidar and camera-based systems. The algorithm can perform inference on the Jetson Xavier AGX at rates of ~20 to ~220 ms depending on the number of objects and their corresponding point amounts in the KITTI dataset.",

keywords = "3D object detection, automated driving, CUDA, sensor data fusion",

author = "Topi Miekkala and Matti Kutila and Mathias Schneider and Alfred H{\"o}{\ss}",

note = "Publisher Copyright: {\textcopyright} 2022 IEEE.; 2022 IEEE 18th International Conference on Intelligent Computer Communication and Processing, ICCP 2022 ; Conference date: 22-09-2022 Through 24-09-2022",

year = "2022",

doi = "10.1109/ICCP56966.2022.10053969",

language = "English",

pages = "113--118",

editor = "Sergiu Nedevschi and Rodica Potolea and Slavescu, {Radu Razvan}",

booktitle = "Proceedings - 2022 IEEE 18th International Conference on Intelligent Computer Communication and Processing Conference, ICCP 2022",

publisher = "IEEE Institute of Electrical and Electronic Engineers",

address = "United States",

}

Miekkala, T , Kutila, M, Schneider, M & Höß, A 2022, Optimizing 3D object detection for embedded systems in automated vehicles using sensor data fusion and CUDA computing. in S Nedevschi, R Potolea & RR Slavescu (eds), Proceedings - 2022 IEEE 18th International Conference on Intelligent Computer Communication and Processing Conference, ICCP 2022. IEEE Institute of Electrical and Electronic Engineers, pp. 113-118, 2022 IEEE 18th International Conference on Intelligent Computer Communication and Processing, ICCP 2022, Cluj-Napoca, Romania, 22/09/22. https://doi.org/10.1109/ICCP56966.2022.10053969

Optimizing 3D object detection for embedded systems in automated vehicles using sensor data fusion and CUDA computing. / Miekkala, Topi ; Kutila, Matti; Schneider, Mathias et al.
Proceedings - 2022 IEEE 18th International Conference on Intelligent Computer Communication and Processing Conference, ICCP 2022. ed. / Sergiu Nedevschi; Rodica Potolea; Radu Razvan Slavescu. IEEE Institute of Electrical and Electronic Engineers, 2022. p. 113-118.

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

TY - GEN

T1 - Optimizing 3D object detection for embedded systems in automated vehicles using sensor data fusion and CUDA computing

AU - Miekkala, Topi

AU - Kutila, Matti

AU - Schneider, Mathias

AU - Höß, Alfred

PY - 2022

Y1 - 2022

N2 - This article explores the utilization of the processing power of GPUs using CUDA computation for real-time aggregation of multi-sensor data and detection of 3D objects using parallel clustering algorithms. The purpose is to implement an algorithm that fuses raw lidar point cloud data and 2D camera image object detections to produce 3D object clusters in a lidar point cloud. Most of the computation has been implemented using CUDA parallelism to investigate the capability of GPU devices in this task, which is a common challenge in automated driving. The results indicate that processing times can be optimized within the algorithm, which is crucial when considering the large amounts of data provided by lidar and camera-based systems. The algorithm can perform inference on the Jetson Xavier AGX at rates of ~20 to ~220 ms depending on the number of objects and their corresponding point amounts in the KITTI dataset.

AB - This article explores the utilization of the processing power of GPUs using CUDA computation for real-time aggregation of multi-sensor data and detection of 3D objects using parallel clustering algorithms. The purpose is to implement an algorithm that fuses raw lidar point cloud data and 2D camera image object detections to produce 3D object clusters in a lidar point cloud. Most of the computation has been implemented using CUDA parallelism to investigate the capability of GPU devices in this task, which is a common challenge in automated driving. The results indicate that processing times can be optimized within the algorithm, which is crucial when considering the large amounts of data provided by lidar and camera-based systems. The algorithm can perform inference on the Jetson Xavier AGX at rates of ~20 to ~220 ms depending on the number of objects and their corresponding point amounts in the KITTI dataset.

KW - 3D object detection

KW - automated driving

KW - CUDA

KW - sensor data fusion

UR - http://www.scopus.com/inward/record.url?scp=85149859269&partnerID=8YFLogxK

U2 - 10.1109/ICCP56966.2022.10053969

DO - 10.1109/ICCP56966.2022.10053969

M3 - Conference article in proceedings

SP - 113

EP - 118

BT - Proceedings - 2022 IEEE 18th International Conference on Intelligent Computer Communication and Processing Conference, ICCP 2022

A2 - Nedevschi, Sergiu

A2 - Potolea, Rodica

A2 - Slavescu, Radu Razvan

PB - IEEE Institute of Electrical and Electronic Engineers

T2 - 2022 IEEE 18th International Conference on Intelligent Computer Communication and Processing, ICCP 2022

Y2 - 22 September 2022 through 24 September 2022

ER -

Miekkala T , Kutila M, Schneider M, Höß A. Optimizing 3D object detection for embedded systems in automated vehicles using sensor data fusion and CUDA computing. In Nedevschi S, Potolea R, Slavescu RR, editors, Proceedings - 2022 IEEE 18th International Conference on Intelligent Computer Communication and Processing Conference, ICCP 2022. IEEE Institute of Electrical and Electronic Engineers. 2022. p. 113-118 doi: 10.1109/ICCP56966.2022.10053969

Optimizing 3D object detection for embedded systems in automated vehicles using sensor data fusion and CUDA computing

Abstract

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this